Peter F. Epstein
Research Statement
In my dissertation, titled Sensible Concepts: Experience and the A Priori, I develop a novel account of spatial
experience that—unlike most contemporary theories of perception—situates our experience of space within a
broader context of non-sensory cognitive activities. On my account, to perceive an object as square is, in part,
to deploy an a priori Euclidean concept of squareness – a concept that features in, but is not derived from,
experience. This broadly rationalist approach to perception provides a framework for fruitful investigation of
many central questions in the philosophy of mind and in metaphysics.
TEMPORAL EXPERIENCE: On the account of spatial perception I develop in my dissertation, our a priori
concepts of space allow us to grasp the nature of the spatial properties our experiences represent. In the case
of the secondary qualities, by contrast, I argue that, lacking any a priori grasp of a property like redness, we can
represent such a property only by way of its role in experience – as whatever property plays the relevant role in
causing experiences of red. Since perception does not inform us which specific property is playing that causal
role, we are left in the dark about the nature of the secondary qualities. But there is another set of properties
represented in perception—temporal properties—which play more than a merely causal role in our experience. The temporal properties represented by our perception of external events are properties our experiences
themselves can instantiate: both a headache and the transit of Venus can last for six hours; both an itch and the
opening of a steam valve can occur at regular thirty-second intervals. This raises the intriguing possibility that
our experiences supply us with concepts of time in virtue of instantiating temporal properties. On this
picture, we can grasp the nature of the temporal properties we represent worldly events as having, in virtue of
the concepts we derive from experiences that instantiate those very properties. Temporal experiences, then,
share features with both our color experiences and our spatial experiences. On the one hand, temporal
experiences, like color experiences, employ concepts that are derived from experience. On the other hand,
temporal experiences, like spatial experiences, deploy concepts that give us insight into the nature of the
properties represented. In future work, I plan to explore the dual experiential role of temporal properties:
they are properties our experiences both represent and instantiate.
CONNECTIONS BETWEEN PRIMARY-QUALITY AND SECONDARY-QUALITY CONTENTS: Another avenue for
future research concerns the distinction between primary and secondary qualities that I elaborate in my
dissertation. There, I emphasize that primary qualities and secondary qualities show up in experience in very
different ways: a primary quality like squareness is represented via an a priori concept, which captures its
essence; while, in the case of a secondary quality, like orange, our experience gives us no insight into the
nature of the property represented. But shape and color do not show up in experience disjointly. The orange
and yellow of a Rothko painting are perceived as spread over areas of canvas with particular shapes. On my
picture, the representational content of such an experience will be opaque in certain respects (it will provide
no insight into the nature of orange) but transparent in others (it will represent the color quality in question—
whatever that quality is—as distributed in a particular spatial manner). The connections here will be complex;
my account of perception provides a framework for illuminating investigation of those complexities.
CONTEMPORARY PHYSICS AND THE INTERPRETATION OF SCIENTIFIC THEORY: In the final chapter of my
dissertation, I address the widely-accepted claim that Einstein’s special theory of relativity (STR) calls into
question the applicability of our determinate a priori spatial concepts to the physical world. I reject that claim,
arguing that careful analysis of STR shows that our perceptions of shape can veridically attribute standard
shape properties to objects in a relativistic universe. But there are two other areas of contemporary science
that might seem to put pressure on the veridicality of our shape perceptions. First, there is Einstein’s general
theory of relativity (GTR), which is standardly interpreted as revealing that the space in which we live is nonEuclidean. Since our a priori spatial concepts are Euclidean, GTR seems to conflict with my claim that we
veridically apply our a priori spatial concepts to the physical objects we perceive. The second challenge stems
from quantum mechanics, which, on some interpretations, leaves no room for space (or space-time) as a real
feature of the physical universe. These challenges are particular instances of a larger question that I aim to
explore within the context of the framework developed in my dissertation: How should we understand the
relationship between the conceptual resources we deploy in our perceptual experience of the world—
resources that are, on my view, largely a priori in origin—and the picture of physical reality we get from
1
Peter F. Epstein
Research Statement
scientific investigation? According to my account, our very possession of a conception of the physical world
depends on our deployment of a priori concepts in perception. But scientific investigation is itself dependent
on perception, and it aims to reveal truths about the physical world. So scientific theories cannot consistently
renounce our a priori conceptual scheme in its entirety; doing so would leave those theories with no coherent
account of the very evidence on which they are based, or of the world they are intended to describe. This
imposes a useful constraint on scientific theorizing: we will have to interpret our scientific theories in a way
that makes room for an intelligible account of our perceptual contact with the empirical world. This constraint can help guide us to a fuller understanding of the discoveries of contemporary physics.
OBJECTIVE REPRESENTATION: The account I develop in my dissertation also provides for a novel approach
to a currently-vexed question: How is objective representation achieved? Tyler Burge has recently argued that
we can understand the phenomenon of “perceptual objectivity” as a primitive feature of the cognitive states
of many organisms, which emerges from various extrinsic relations between an organism and its environment. 1 This perspective, Burge claims, stands strongly opposed to a dialectic that dominated much of
twentieth-century philosophy, according to which we need to understand how it is possible for an individual to
have a conceptual scheme that allows for objective representation. On my account of perceptual representation,
we can see why these two approaches, to a large extent, talk past each other. Burge-style externalist accounts
of representation can provide a story about a kind of objectivity possessed by an organism’s perceptual states;
but that story does not itself make intelligible how a subject can have the concepts of mind-independence that
must be part of any full account of our perceptual experience. Our experience does not just represent
objective features of the mind-independent world; the mind-independence of the features represented is
something that we consciously grasp in having the experiences we do. This latter feature of experience is left
out of Burge’s picture; but it is the central motivation for the twentieth-century narrative, on which an
individual’s understanding of mind-independence is the explanatory target. My account of perception can shed
light on this latter question, by emphasizing the way that, in addition to any externally-determined objective
content our perceptual states might have (the contents that are Burge’s focus, which we might indeed share
with non-rational animals), those states are also ones that we understand as representations of mindindependent objects, in virtue of our a priori grasp of spatial properties.
FORMAL EPISTEMOLOGY AND DECISION THEORY
During my doctoral studies, I have pursued a secondary research project focusing on foundational issues in
formal epistemology and decision theory. These fields, I argue, suffer from inadequate attention to the
relation between their mathematically-articulated theories and our intuitive understanding of the phenomena
analyzed. Our commonsense concepts of mental states like belief and preference provide the only possible
starting point for the development of a precise formal theory of rationality. So a firm understanding of those
mental states is a prerequisite for any such formal theory. In my research, I investigate the nature of rational
decision-making and scientific discovery through the lens of my background in philosophy of mind. This
approach opens up a novel perspective on questions typically addressed through more purely formalistic
methods. Below, I describe some current work and directions for future research related to this project.
UTILITY IN DECISION THEORY: In a paper titled “Against Operationalism” (a version of which I presented
at the 2011 Formal Epistemology Workshop), I argue for a realist interpretation of the decision theoretic
notion of utility. Anti-realism or “operationalism” about utility—which is the orthodox view amongst decision
theorists—denies that the notion of utility that features in decision theory represents any real psychological
phenomenon. Instead, on the operationalist interpretation, utility is just a convenient way to represent an
agent’s choice behavior. Such a picture, I argue—by divorcing the notion of utility from our pre-theoretical
conception of the strengths of our preferences—leaves the axioms at decision theory’s foundation without their
needed mooring. If decision theory is to serve as a theory of practical rationality, its axioms must be defensi-
1
Burge, Tyler (2010). Origins of Objectivity. Oxford University Press.
2
Peter F. Epstein
Research Statement
ble as rational requirements. This defense, in turn, is possible only if we understand the mathematicallyprecise concept of utility employed in the theory as deriving from our commonsense understanding of the
preferences that drive us to act. And those preferences, given the kinds of agents we are, come in degrees of
strength. Thus, we need to embrace a realist conception of utility as a measure of the strength of our preferences in order to explain how decision theory can provide normative guidance for us as human agents.
SELF-LOCATING BELIEF: Two very different kinds of mental states are often lumped together in formal
epistemology under the label “credence.” The first is the kind of state an agent is in when she believes that a
certain process or event has some objective likelihood of occurring. This is the kind of state we are in when
we believe that a coin has a 50% chance of landing heads. The second kind of state occurs when an agent
simply lacks relevant information – when she is ignorant about some fact. In some instances—particularly
those involving self-locating belief—this second kind of state tracks propositions that cannot intelligibly be
taken to involve any kind of “probabilistic process” at all. For example, I may not know whether it is Monday
or Tuesday; but that is not the same as my believing that some random process, like the tossing of a coin, has
somehow “determined what day it is.” Thus, classifying both kinds of states simply as “credences” masks
important features of those states. In a paper titled “Sleeping Beauty and the Principal Principle,” I show that
the standard argument for the “1/3” answer in the Sleeping Beauty case ignores this distinction between the
two kinds of mental states that are classified as “credences.” Thus, the argument is guilty of a kind of equivocation: it treats two very different senses of “likelihood”—the objective chances we associate with physical
processes and the uncertainty an agent might have about her own situation—as equivalent when applying
formal methods of analysis. I offer an alternative analysis, based on my explication of the different kinds of
mental states we can be in when we lack certainty about some proposition. This analysis reveals that we
should not let the mere ignorance embodied in self-locating uncertainty undermine the contentful judgments
about probabilistic processes embodied in our assessments of objective chances. As a result, we should accept
the ½ answer in the Sleeping Beauty problem.
BAYESIAN CONFIRMATION THEORY: It is typically assumed that, in applying the standard Bayesian rule for
assessing hypotheses in the light of evidence, we ought to reason on the basis of all of the available information. I argue that, surprisingly, this “requirement of total evidence” fails to deliver the correct verdicts
when applied to simple scientific experiments. In analyzing the evidential import of a given observation, we
must not be overly-specific in our classification of that observation. To see this, consider the following case.
You know there are some large fish and some small fish in a lake, but you don’t know the proportions: either
there are mostly large fish (HL) or mostly small fish (HS). You know that you will catch one fish each time you
stick in your net. You stick in your net and catch a large fish. This would seem to be evidence in favor of HL
over HS, since the observed outcome—the catching of a large fish—was more likely to occur if the fish in the
lake are mostly large, rather than mostly small. But you could characterize your observation in a more specific
way: the large fish you caught is a particular large fish. If you name her “Asha,” then the most-specific way to
classify your observation is: You caught the large fish Asha. But now it is unclear why your observation
counts in favor of HL: you were no more likely to catch Asha, specifically, if more of the other fish are large,
rather than small. By classifying your observation in the most-specific possible way, you mask the confirmation your experimental procedure in fact supplied. This puzzling phenomenon raises important questions
about how to model sound reasoning about probabilistic processes. In a paper titled “The Fine-Tuning
Argument and the Requirement of Total Evidence” (forthcoming in Philosophy of Science), I explore these
questions in the context of the so-called “fine-tuning argument” for the existence of multiple universes. I
argue that, in assessing questions about the process that gave rise to our own universe, the possible outcomes
of that process should be partitioned coarsely, without reference to our own unique perspective as the beings
who actually resulted from it. Thinking in this way avoids the kind of post hoc assessment of evidence we
recognize as problematic in the more mundane case of the fish-catching experiment.
3